Clickjacking Revisited A Perceptual View of UI Security Devdatta - - PowerPoint PPT Presentation

clickjacking revisited
SMART_READER_LITE
LIVE PREVIEW

Clickjacking Revisited A Perceptual View of UI Security Devdatta - - PowerPoint PPT Presentation

Clickjacking Revisited A Perceptual View of UI Security Devdatta Akhawe / Warren He / Zhiwei Li/ Reza Moazzezi / Dawn Song University of California, Berkeley Clickjacking is a malicious technique of tricking a Web user into clicking on something


slide-1
SLIDE 1

Clickjacking Revisited

A Perceptual View of UI Security

Devdatta Akhawe / Warren He / Zhiwei Li/ Reza Moazzezi / Dawn Song University of California, Berkeley

slide-2
SLIDE 2

Clickjacking is a malicious technique of tricking a Web user into clicking on something different from what the user perceives

(wikipedia)

slide-3
SLIDE 3

Five novel clickjacking attacks that bypass current defenses Evaluation with 250 users on MTurk Today

slide-4
SLIDE 4

Attack Setup

  • Attacker wants to trick user into clicking a button, in
  • ur case, the Facebook like button
  • Attacker convinces user to play a game on attacker

controlled webpage

  • Attacker can frame the Facebook Like button, but has

no control over the FB display area/frame

  • Attacker has full control of remaining display area
slide-5
SLIDE 5

Attacker page

slide-6
SLIDE 6

A successful attack (bypassing current defenses) requires the like button be fully visible for a noticeable amount of time (say ~500ms)

slide-7
SLIDE 7

Destabilizing Pointer Perception

slide-8
SLIDE 8

Video Demo

slide-9
SLIDE 9

Player starts moving mouse Fake pointer starts moving to the left (red) Finally, close to the target, player corrects in a sudden motion, moving the real pointer towards right Player clicks Like button by mistake User keeps moving up and right (black), but fake pointer (red) stays left, confusing the user

slide-10
SLIDE 10

Successful Attack

  • One concern is the appearance of the real

pointer when it approaches the like button

– Attacker has no control over “Like” button frame

  • Key Idea: distract the player’s attention with

lots of moving images

slide-11
SLIDE 11

Real Attack

slide-12
SLIDE 12

Attacking Peripheral Vision

slide-13
SLIDE 13

Game Setup

slide-14
SLIDE 14

Player must leave mouse at bottom

  • f screen

But, watch main game area at top right

slide-15
SLIDE 15

(a) time Sensor Blocks Player

slide-16
SLIDE 16

pause (c) (b) (a) time Sensor Blocks Player

slide-17
SLIDE 17

Motor Adaptation

slide-18
SLIDE 18

Game Setup

slide-19
SLIDE 19

Player presented with asteroid

slide-20
SLIDE 20

Asteroid explodes when clicked Mineral produced at constant displacement Player must click

  • n this mineral

for points

slide-21
SLIDE 21

Once trained, put like button instead of mineral

slide-22
SLIDE 22

Fast Motion Mislocalization

slide-23
SLIDE 23

Game Setup

slide-24
SLIDE 24

Player presented with asteroid with spinning arrow

slide-25
SLIDE 25

When arrow stop, mineral shoots out Player must click on mineral for points

slide-26
SLIDE 26

The Flash Lag Effect

  • Flash lag is a visual illusion where a moving
  • bject, at a particular instant, seems further

ahead than it actually is

  • Brain predicts future displacement
  • The player’s click is actually beyond the

mineral, but we still award points

slide-27
SLIDE 27

After a few trials, put like button beyond mineral

slide-28
SLIDE 28

Visual Cues and Click Timing

slide-29
SLIDE 29

Game Setup

slide-30
SLIDE 30

Negative points for clicking on grey asteroid Positive points for clicking on red asteroid

slide-31
SLIDE 31

Move asteroid under a like button

slide-32
SLIDE 32

Evaluation

slide-33
SLIDE 33

Evaluation

Attack Name Number of subjects Success Rate (%) Destabilizing Pointer Perception 50 100 Peripheral Vision 49 51.02 Adaptation 46 28.26 Fast Motion Mislocalization 47 27.66 Visual Cue for timing 50 50

  • MTurk study with 50 workers for each attack.
  • Some subjects exited before completing the

exercise Attacks 2 through 5 work for touch devices too!

slide-34
SLIDE 34

This is only a lower-bound …

slide-35
SLIDE 35
  • Our attacks are simple. Possible to

dynamically adapt the attack as user plays the games.

  • Better models of pointer movement and click

prediction can improve success rates.

  • Each attack targets a different limitation of

human perception. A combined attack likely to achieve 100% success.

Complex Attacks

slide-36
SLIDE 36
  • Human perception is a vast and well studied
  • topic. Many more attacks possible.
  • For example, Change Blindness:

– Well studied phenomenon in which user fails to notice difference in two images. – Attacker can switch in a like button and an appropriately primed user won’t notice.

More Attacks

slide-37
SLIDE 37

Future Work

slide-38
SLIDE 38

Future Work

  • Secure UI design needs to take human

perception in account while designing interfaces

– Changes needed to specifications such as the UI Security specification

  • Computer Vision based techniques (or machine

perception) could be key for defenses

  • Designing a secure user interaction mechanism

critical for security

slide-39
SLIDE 39

evil@berkeley.edu

questions?